DeepACC：融合先验知识的深度学习框架基于中期图像自动进行染色体分类
计算机视觉与模式识别
2021-08-18 v2
摘要
染色体分类是核型分析中的重要但困难且繁琐的任务。以往方法仅对人工分割的单个染色体进行分类，远未满足临床实践。本工作中，我们提出一种基于检测的方法DeepACC，基于整张中期图像同时定位并精细分类染色体。我们首先引入加性角间隔损失以增强模型的判别力。为缓解批次效应，我们通过一个充分利用染色体通常成对出现的先验知识的孪生网络，逐例变换每类的决策边界。此外，我们将临床七组判据作为先验知识，设计额外的组内邻接损失以进一步降低类间相似性。我们从临床实验室收集并标注了3390张中期图像以评估性能。结果表明，与最先进的基线相比，新设计带来了令人鼓舞的性能提升。
引用
@article{arxiv.2006.15528,
title = {DeepACC:Automate Chromosome Classification based on Metaphase Images using Deep Learning Framework Fused with Prior Knowledge},
author = {Chunlong Luo and Tianqi Yu and Yufan Luo and Manqing Wang and Fuhai Yu and Yinhao Li and Chan Tian and Jie Qiao and Li Xiao},
journal= {arXiv preprint arXiv:2006.15528},
year = {2021}
}
备注
This work is supported by a fund from another hospital. Only Li Xiao conceived the idea and supervised Chunlong Luo to complete the work, the data provider did not participate in the research process. Thus, the authorships and institutional information are not correct. After careful consideration, I decide to withdraw this preprint version